A COLLECTION OF RAW DATA, taken by itself, is no more exciting or informative than junk
mail before Election Day. Whether you have neatly arranged the data in rows on a data col-
lection form or scribbled them on the back of an out-of-date announcement you tore from
the bulletin board, a collection of numbers is still just a collection of numbers. To be inter-
pretable, they first must be organized in some sort of logical order. The following actual
experiment illustrates some of these steps.
How do human beings process information that is stored in their short-term memory? If I
asked you to tell me whether the number “6” was included as one of a set of five digits that
you just saw presented on a screen, do you use sequential processing to search your short-term
memory of the screen and say “Nope, it wasn’t the first digit; nope, it wasn’t the second,” and
so on? Or do you use parallel processing to compare the digit “6” with your memory of all the
previous digits at the same time? The latter approach would be faster and more efficient, but
human beings don’t always do things in the fastest and most efficient manner. How do you
think that you do it? How do you search back through your memory and identify the person
who just walked in as Jennifer? Do you compare her one at a time with all the women her age
whom you have met, or do you make comparisons in parallel? (This second example uses
long-term memory rather than short-term memory, but the questions are analogous.)
In 1966, Sternberg ran a simple, famous, and important study that examined how people
recall data from short-term memory. This study is still widely cited in the research literature.
On a screen in front of the subject, he briefly presented a comparison set of one, three, or
five digits. Shortly after each presentation he flashed a single test digit on the screen and re-
quired the subject to push one button (the positive button) if the test digit had been included
in the comparison set or another button (the negative button) if the test digit had not been
part of the comparison set. For example, the two stimuli might look like this:Comparison 27481
Test 5(Remember, the two sets of stimuli were presented sequentially, not simultaneously, so only
one of those lines was visible at a time.) The numeral “5” was not part of the comparison set,
and the subject should have responded by pressing the negative button. Sternberg measured
the time, in 100ths of a second, that the subject took to respond. This process was repeated
over many randomly organized trials. Because Sternberg was interested in how people
process information, he was interested in how reaction times varied as a function of the num-
ber of digits in the comparison set and as a function of whether the test digit was a positive or
negative instance for that set. (If you make comparisons sequentially, the time to make a de-
cision should increase as the number of digits in the comparison set increases. If you make
comparisons in parallel, the number of digits in the comparison set shouldn’t matter.)
Although Sternberg’s goal was to compare data for the different conditions, we can
gain an immediate impression of our data by taking the full set of reaction times, regard-
less of the stimulus condition. The data in Table 2.1 were collected in an experiment simi-
lar to Sternberg’s but with only one subject—myself. No correction of responses was
allowed, and the data presented here come only from correct trials.2.1 Plotting Data
As you can see, there are simply too many numbers in Table 2.1 for us to be able to inter-
pret them at a glance. One of the simplest methods to reorganize data to make them more
intelligible is to plot them in some sort of graphical form. There are several common ways16 Chapter 2 Describing and Exploring Data